2 $Id: marcomp.c,v 1.1 2003-02-28 12:33:39 oleg Exp $
4 marcomp.c - compiler of MARC statements.
11 #include <yaz/yaz-util.h>
15 static mc_token mc_gettoken(mc_context *c);
16 static void mc_ungettoken(mc_context *c);
17 static int mc_getval(mc_context *c);
18 static int mc_getdata(mc_context *c, char *s, int sz);
19 static void mc_getinterval(mc_context *c, int *start, int *end);
21 static mc_subfield *mc_mk_subfield(mc_subfield *parent);
22 static mc_field *mc_mk_field(void);
24 static struct mc_errmsg
31 {EMCF, "not complete field"},
32 {EMCSF, "not complete subfield"},
33 {EMCSFGROUP, "not closed GROUP"},
34 {EMCSFVAR, "not closed VARIANT"},
35 {EMCSFINLINE, "not closed IN-LINE"},
36 {EMCEND, "not correct errno"}
38 mc_errcode mc_errno(mc_context *c)
42 const char *mc_error(mc_errcode no)
44 if (no >= EMCOK && no<EMCEND)
45 return mc_errmsg[no].msg;
47 return mc_errmsg[EMCEND].msg;
49 mc_context *mc_mk_context(const char *s)
55 p = (mc_context*) xmalloc(sizeof(*p));
60 memset(p, 0, sizeof(*p));
69 void mc_destroy_context(mc_context *c)
73 mc_token mc_gettoken(mc_context *c)
75 if (c->offset >= c->len)
78 switch (*(c->data+c->offset))
80 case '{': c->crrtok = LVARIANT; break;
81 case '}': c->crrtok = RVARIANT; break;
82 case '(': c->crrtok = LGROUP; break;
83 case ')': c->crrtok = RGROUP; break;
84 case '<': c->crrtok = LINLINE; break;
85 case '>': c->crrtok = RINLINE; break;
86 case '$': c->crrtok = SUBFIELD; break;
87 case '[': c->crrtok = LINTERVAL; break;
88 case ']': c->crrtok = RINTERVAL; break;
90 if (isspace(*(c->data+c->offset)) || *(c->data+c->offset) == '\n')
97 c->crrval = *(c->data+c->offset);
101 fprintf(stderr, "gettoken(): offset: %d", c->offset);
102 if (c->crrtok == REGULAR)
103 fprintf(stderr, "<%c>", c->crrval);
104 fprintf(stderr, "\n");
109 void mc_ungettoken(mc_context *c)
114 int mc_getval(mc_context *c)
118 int mc_getdata(mc_context *c, char *s, int sz)
124 if (mc_gettoken(c)!=REGULAR)
135 void mc_getinterval(mc_context *c, int *start, int *end)
138 int start_pos, end_pos;
140 start_pos = end_pos = -1;
142 if (mc_gettoken(c) == LINTERVAL)
148 mc_token tok = mc_gettoken(c);
150 if (tok == RINTERVAL || tok == NOP)
153 buf[i] = mc_getval(c);
157 i = sscanf(buf, "%d-%d", &start_pos, &end_pos);
169 mc_field *mc_mk_field(void)
171 mc_field *p = (mc_field *)xmalloc(sizeof(*p));
175 memset(p, 0, sizeof(*p));
176 p->name = (char *)xmalloc(SZ_FNAME+1);
178 p->ind1 = (char *)xmalloc(SZ_IND+1);
180 p->ind2 = (char *)xmalloc(SZ_IND+1);
182 p->interval.start = p->interval.end = -1;
186 void mc_destroy_field(mc_field *p)
190 if (p->name) xfree(p->name);
191 if (p->ind1) xfree(p->ind1);
192 if (p->ind2) xfree(p->ind2);
193 if (p->list) mc_destroy_subfields_recursive(p->list);
196 mc_field *mc_getfield(mc_context *c)
204 c->errcode = EMCNOMEM;
208 if (mc_getdata(c, pf->name, SZ_FNAME) == SZ_FNAME)
210 mc_token nexttok = mc_gettoken(c);
214 if (nexttok == LINTERVAL)
216 mc_getinterval(c, &pf->interval.start, &pf->interval.end);
218 fprintf(stderr, "ineterval (%d)-(%d)\n", pf->interval.start,
223 if ((mc_getdata(c, pf->ind1, SZ_IND) == SZ_IND) &&
224 (mc_getdata(c, pf->ind2, SZ_IND) == SZ_IND))
226 pf->list = mc_getsubfields(c, 0);
232 mc_destroy_field(pf);
238 mc_subfield *mc_mk_subfield(mc_subfield *parent)
240 mc_subfield *p = (mc_subfield*)xmalloc(sizeof(*p));
244 memset(p, 0, sizeof(*p));
246 p->name = (char *)xmalloc(SZ_SFNAME+1);
248 p->prefix = (char *)xmalloc(SZ_PREFIX+1);
250 p->suffix = (char *)xmalloc(SZ_SUFFIX+1);
253 p->interval.start = p->interval.end = -1;
257 void mc_destroy_subfield(mc_subfield *p)
262 if (p->which == MC_SFGROUP || p->which == MC_SFVARIANT)
265 mc_destroy_subfields_recursive(p->u.child);
267 else if (p->which == MC_SF)
270 mc_destroy_field(p->u.in_line);
272 if (p->name) xfree(p->name);
273 if (p->prefix) xfree(p->prefix);
274 if (p->suffix) xfree(p->suffix);
275 if (p->parent) p->parent->next = p->next;
278 void mc_destroy_subfields_recursive(mc_subfield *p)
283 mc_destroy_subfields_recursive(p->next);
285 if (p->which == MC_SFGROUP || p->which == MC_SFVARIANT)
288 mc_destroy_subfields_recursive(p->u.child);
290 else if (p->which == MC_SF)
293 mc_destroy_field(p->u.in_line);
296 if (p->name) xfree(p->name);
297 if (p->prefix) xfree(p->prefix);
298 if (p->suffix) xfree(p->suffix);
299 if (p->parent) p->parent->next = 0;
302 mc_subfield *mc_getsubfields(mc_context *c, mc_subfield *parent)
305 mc_token tok = mc_gettoken(c);
312 if (!(psf = mc_mk_subfield(parent)))
314 c->errcode = EMCNOMEM;
318 psf->which = MC_SFGROUP;
319 psf->u.child = mc_getsubfields(c, psf);
321 if (mc_gettoken(c) == RGROUP)
322 psf->next = mc_getsubfields(c, psf);
325 c->errcode = EMCSFGROUP;
326 mc_destroy_subfield(psf);
330 else if (tok == LVARIANT)
332 if (!(psf = mc_mk_subfield(parent)))
334 c->errcode = EMCNOMEM;
338 psf->which = MC_SFVARIANT;
339 psf->u.child = mc_getsubfields(c, psf);
341 if (mc_gettoken(c) == RVARIANT)
342 psf->next = mc_getsubfields(c, psf);
345 c->errcode = EMCSFVAR;
346 mc_destroy_subfield(psf);
350 else if (tok == RGROUP || tok == RVARIANT || tok == RINLINE)
355 else if (tok == REGULAR)
357 if (!(psf = mc_mk_subfield(parent)))
359 c->errcode = EMCNOMEM;
365 if((mc_getdata(c, psf->prefix, SZ_PREFIX) == SZ_PREFIX) &&
366 (mc_gettoken(c) == SUBFIELD) &&
367 (mc_getdata(c, psf->name, SZ_SFNAME) == SZ_SFNAME))
369 mc_token tok = mc_gettoken(c);
373 if (tok == LINTERVAL)
375 mc_getinterval(c, &psf->interval.start, &psf->interval.end);
377 else if (tok == LINLINE)
380 psf->u.in_line = mc_getfield(c);
381 if (mc_gettoken(c) != RINLINE)
383 c->errcode = EMCSFINLINE;
384 mc_destroy_subfield(psf);
389 if (mc_getdata(c, psf->suffix, SZ_SUFFIX) == SZ_SUFFIX)
392 psf->next = mc_getsubfields(c, psf);
397 mc_destroy_subfield(psf);